Can I A/B test a Steam store page?

Not with a built-in tool. The Steamworks documentation describes no way to show two versions of a page to two groups of players. Every visitor sees the same page. So you test in one of two ways: compare two periods with different pages, or compare two groups of outside traffic.

Each way has a weak spot. Periods differ in sales, festivals and news. Outside traffic differs from Steam traffic. Use both ways together when the result matters.

MethodWhat changesWhat you compareMain risk
Two periodsThe page itself, such as the capsule or trailerClick-through rate and wishlist adds per visit, week against weekA sale or festival in one week skews it
Two tagged linksThe message or audience before the clickWishlists per tracked visit for each utm_contentTests the ad or post, not the page
Two tagged links plus two periodsThe page, with fixed outside trafficWishlists per tracked visit on the same link, before and afterNeeds steady outside traffic

Sources: Steamworks UTM Analytics documentation; Steamworks Store and Platform Traffic Reporting documentation.

How do I test a page change with two periods?

Change one thing on the page, then compare a calm week before with a calm week after. A calm week has no festival, no sale on your game and no big press hit. Changing two things at once tells you nothing about either.

Read the click-through rate in the traffic breakdown for capsule changes. Valve defines it as the share of impressions that result in visits. Read wishlist adds divided by visits for changes on the page, such as the trailer or the screenshots.

  1. Write down the change and the date in your marketing log.
  2. Record one calm week before the change: impressions, visits, click-through rate and wishlist adds.
  3. Make one change and publish it.
  4. Wait one full week. Skip weeks with a festival or a sale.
  5. Record the same numbers for the week after and compare the rates, not the totals.
  6. Keep the change only if the rate moved more than your normal week to week swing.

Sources: Steamworks Store and Platform Traffic Reporting documentation; Steamworks Wishlists documentation.

Which metric decides the winner?

Wishlists per tracked visit. Tracked visits are visits by people signed in to Steam, and only they can convert. Total visits include people who never could. Purchases are rare before launch, so wishlists are the better signal.

Wait 4 days after the last visit before you read the result. Conversions are not final until then. Steam also hides UTM combinations with few visits, so a test with tiny traffic may show no rows.

Sources: Steamworks UTM Analytics documentation.

How much traffic does a test need?

More than most indie tests get. Here is arithmetic, not data: if one variant turns 10% of tracked visits into wishlists and the other 12%, then 200 tracked visits each give about 20 and 24 wishlists. A difference of 4 wishlists can come from chance alone.

Our own rule: decide on the size of the test before you start, and do not stop early because one variant leads. If the gap is small, call it a tie and keep the simpler version. A real winner tends to stay ahead when you run the test again.

Sources: Steamworks UTM Analytics documentation.

What should I test first?

Test what most players see. On Steam that is the capsule, because it shows in every list, and the first trailer. Outside Steam, test the hook in the first line of the post or ad.

Do not over-trust click-through rate. In a survey of over 100 games, How To Market A Game found no strong link between click-through rate and wishlists earned. A capsule that gets more clicks from the wrong players can lower wishlists per visit.

  • Capsule art: genre clear at small size, or not.
  • Trailer: gameplay in the first seconds, or story first.
  • Short description: the player verb first, or the setting first.
  • Ad or post hook: a feature, a question or a clip.

Sources: How To Market A Game (14 September 2022); Steamworks Store and Platform Traffic Reporting documentation.

What are the limits of UTM testing?

Attribution breaks when a player changes browser or device, or opens the Steam client and searches for the game. Social app browsers make it hard to act on Steam, so many actions go untracked. Some players are not tracked because of cookie settings. So both variants miss part of their wishlists. The comparison is fair only if both variants lose about the same part, which is true when the traffic sources are alike.

Sources: Steamworks UTM Analytics documentation.

How do I run the test without spreadsheets?

Manalens is the analytics and growth engine built for games. Create two tracked short links with the same target and a different utm_content. Each click records an event, so you see clicks per variant. Import the UTM report and group the Steam page by content to read wishlists per tracked visit for each variant.

Sources: Manalens product documentation.

Frequently asked questions

Does Steam have built-in A/B testing for store pages?

The Steamworks documentation describes none. Every visitor sees the same page. Test with two periods, before and after one change, or split your own outside traffic with two tagged links and compare wishlists per tracked visit in the UTM Analytics report.

Which UTM field should I use for variants?

Use utm_content, such as variant_a and variant_b, and keep utm_source and utm_campaign the same. Steam reads all five UTM fields, and every link needs utm_source or utm_campaign. Keep the number of values small, because Steam hides UTM combinations with few visits from the report.

How long should a Steam page test run?

At least one full week per variant or period, on calm weeks with no festival or sale. Then wait 4 days after the last visit, because Steam finalizes UTM conversions 4 days after the visit. Small games often need several weeks to see a real gap.

Should I pick the capsule with the higher click-through rate?

Not on that alone. A 2022 How To Market A Game survey found no strong link between click-through rate and wishlists. Check wishlist adds per visit too, because a capsule that draws the wrong players can raise clicks and lower wishlists.

Can I test two Steam capsules at the same time?

Not on Steam itself, because every visitor sees the same capsule. You can test two images in your own ads or posts, each with its own tagged link, and compare wishlists per tracked visit. Then put the winner on Steam and check the click-through rate over two calm weeks.

Sources and implementation